src/librustc_mir/interpret/machine.rs

   1 //! This module contains everything needed to instantiate an interpreter.
   2 //! This separation exists to ensure that no fancy miri features like
   3 //! interpreting common C functions leak into CTFE.
   4
   5 use std::borrow::{Borrow, Cow};
   6 use std::hash::Hash;
   7
   8 use rustc_middle::mir;
   9 use rustc_middle::ty::{self, Ty};
  10 use rustc_span::def_id::DefId;
  11
  12 use super::{
  13     AllocId, Allocation, AllocationExtra, Frame, ImmTy, InterpCx, InterpResult, Memory, MemoryKind,
  14     OpTy, Operand, PlaceTy, Pointer, Scalar,
  15 };
  16
  17 /// Data returned by Machine::stack_pop,
  18 /// to provide further control over the popping of the stack frame
  19 #[derive(Eq, PartialEq, Debug, Copy, Clone)]
  20 pub enum StackPopJump {
  21     /// Indicates that no special handling should be
  22     /// done - we'll either return normally or unwind
  23     /// based on the terminator for the function
  24     /// we're leaving.
  25     Normal,
  26
  27     /// Indicates that we should *not* jump to the return/unwind address, as the callback already
  28     /// took care of everything.
  29     NoJump,
  30 }
  31
  32 /// Whether this kind of memory is allowed to leak
  33 pub trait MayLeak: Copy {
  34     fn may_leak(self) -> bool;
  35 }
  36
  37 /// The functionality needed by memory to manage its allocations
  38 pub trait AllocMap<K: Hash + Eq, V> {
  39     /// Tests if the map contains the given key.
  40     /// Deliberately takes `&mut` because that is sufficient, and some implementations
  41     /// can be more efficient then (using `RefCell::get_mut`).
  42     fn contains_key<Q: ?Sized + Hash + Eq>(&mut self, k: &Q) -> bool
  43     where
  44         K: Borrow<Q>;
  45
  46     /// Inserts a new entry into the map.
  47     fn insert(&mut self, k: K, v: V) -> Option<V>;
  48
  49     /// Removes an entry from the map.
  50     fn remove<Q: ?Sized + Hash + Eq>(&mut self, k: &Q) -> Option<V>
  51     where
  52         K: Borrow<Q>;
  53
  54     /// Returns data based on the keys and values in the map.
  55     fn filter_map_collect<T>(&self, f: impl FnMut(&K, &V) -> Option<T>) -> Vec<T>;
  56
  57     /// Returns a reference to entry `k`. If no such entry exists, call
  58     /// `vacant` and either forward its error, or add its result to the map
  59     /// and return a reference to *that*.
  60     fn get_or<E>(&self, k: K, vacant: impl FnOnce() -> Result<V, E>) -> Result<&V, E>;
  61
  62     /// Returns a mutable reference to entry `k`. If no such entry exists, call
  63     /// `vacant` and either forward its error, or add its result to the map
  64     /// and return a reference to *that*.
  65     fn get_mut_or<E>(&mut self, k: K, vacant: impl FnOnce() -> Result<V, E>) -> Result<&mut V, E>;
  66
  67     /// Read-only lookup.
  68     fn get(&self, k: K) -> Option<&V> {
  69         self.get_or(k, || Err(())).ok()
  70     }
  71
  72     /// Mutable lookup.
  73     fn get_mut(&mut self, k: K) -> Option<&mut V> {
  74         self.get_mut_or(k, || Err(())).ok()
  75     }
  76 }
  77
  78 /// Methods of this trait signifies a point where CTFE evaluation would fail
  79 /// and some use case dependent behaviour can instead be applied.
  80 pub trait Machine<'mir, 'tcx>: Sized {
  81     /// Additional memory kinds a machine wishes to distinguish from the builtin ones
  82     type MemoryKind: ::std::fmt::Debug + ::std::fmt::Display + MayLeak + Eq + 'static;
  83
  84     /// Tag tracked alongside every pointer. This is used to implement "Stacked Borrows"
  85     /// <https://www.ralfj.de/blog/2018/08/07/stacked-borrows.html>.
  86     /// The `default()` is used for pointers to consts, statics, vtables and functions.
  87     type PointerTag: ::std::fmt::Debug + Copy + Eq + Hash + 'static;
  88
  89     /// Machines can define extra (non-instance) things that represent values of function pointers.
  90     /// For example, Miri uses this to return a function pointer from `dlsym`
  91     /// that can later be called to execute the right thing.
  92     type ExtraFnVal: ::std::fmt::Debug + Copy;
  93
  94     /// Extra data stored in every call frame.
  95     type FrameExtra;
  96
  97     /// Extra data stored in memory. A reference to this is available when `AllocExtra`
  98     /// gets initialized, so you can e.g., have an `Rc` here if there is global state you
  99     /// need access to in the `AllocExtra` hooks.
 100     type MemoryExtra;
 101
 102     /// Extra data stored in every allocation.
 103     type AllocExtra: AllocationExtra<Self::PointerTag> + 'static;
 104
 105     /// Memory's allocation map
 106     type MemoryMap: AllocMap<
 107             AllocId,
 108             (MemoryKind<Self::MemoryKind>, Allocation<Self::PointerTag, Self::AllocExtra>),
 109         > + Default
 110         + Clone;
 111
 112     /// The memory kind to use for copied global memory (held in `tcx`) --
 113     /// or None if such memory should not be mutated and thus any such attempt will cause
 114     /// a `ModifiedStatic` error to be raised.
 115     /// Statics are copied under two circumstances: When they are mutated, and when
 116     /// `tag_allocation` (see below) returns an owned allocation
 117     /// that is added to the memory so that the work is not done twice.
 118     const GLOBAL_KIND: Option<Self::MemoryKind>;
 119
 120     /// Whether memory accesses should be alignment-checked.
 121     fn enforce_alignment(memory_extra: &Self::MemoryExtra) -> bool;
 122
 123     /// Whether to enforce the validity invariant
 124     fn enforce_validity(ecx: &InterpCx<'mir, 'tcx, Self>) -> bool;
 125
 126     /// Entry point to all function calls.
 127     ///
 128     /// Returns either the mir to use for the call, or `None` if execution should
 129     /// just proceed (which usually means this hook did all the work that the
 130     /// called function should usually have done). In the latter case, it is
 131     /// this hook's responsibility to advance the instruction pointer!
 132     /// (This is to support functions like `__rust_maybe_catch_panic` that neither find a MIR
 133     /// nor just jump to `ret`, but instead push their own stack frame.)
 134     /// Passing `dest`and `ret` in the same `Option` proved very annoying when only one of them
 135     /// was used.
 136     fn find_mir_or_eval_fn(
 137         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 138         instance: ty::Instance<'tcx>,
 139         args: &[OpTy<'tcx, Self::PointerTag>],
 140         ret: Option<(PlaceTy<'tcx, Self::PointerTag>, mir::BasicBlock)>,
 141         unwind: Option<mir::BasicBlock>,
 142     ) -> InterpResult<'tcx, Option<&'mir mir::Body<'tcx>>>;
 143
 144     /// Execute `fn_val`.  It is the hook's responsibility to advance the instruction
 145     /// pointer as appropriate.
 146     fn call_extra_fn(
 147         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 148         fn_val: Self::ExtraFnVal,
 149         args: &[OpTy<'tcx, Self::PointerTag>],
 150         ret: Option<(PlaceTy<'tcx, Self::PointerTag>, mir::BasicBlock)>,
 151         unwind: Option<mir::BasicBlock>,
 152     ) -> InterpResult<'tcx>;
 153
 154     /// Directly process an intrinsic without pushing a stack frame. It is the hook's
 155     /// responsibility to advance the instruction pointer as appropriate.
 156     fn call_intrinsic(
 157         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 158         instance: ty::Instance<'tcx>,
 159         args: &[OpTy<'tcx, Self::PointerTag>],
 160         ret: Option<(PlaceTy<'tcx, Self::PointerTag>, mir::BasicBlock)>,
 161         unwind: Option<mir::BasicBlock>,
 162     ) -> InterpResult<'tcx>;
 163
 164     /// Called to evaluate `Assert` MIR terminators that trigger a panic.
 165     fn assert_panic(
 166         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 167         msg: &mir::AssertMessage<'tcx>,
 168         unwind: Option<mir::BasicBlock>,
 169     ) -> InterpResult<'tcx>;
 170
 171     /// Called to evaluate `Abort` MIR terminator.
 172     fn abort(_ecx: &mut InterpCx<'mir, 'tcx, Self>) -> InterpResult<'tcx, !> {
 173         throw_unsup_format!("aborting execution is not supported")
 174     }
 175
 176     /// Called for all binary operations where the LHS has pointer type.
 177     ///
 178     /// Returns a (value, overflowed) pair if the operation succeeded
 179     fn binary_ptr_op(
 180         ecx: &InterpCx<'mir, 'tcx, Self>,
 181         bin_op: mir::BinOp,
 182         left: ImmTy<'tcx, Self::PointerTag>,
 183         right: ImmTy<'tcx, Self::PointerTag>,
 184     ) -> InterpResult<'tcx, (Scalar<Self::PointerTag>, bool, Ty<'tcx>)>;
 185
 186     /// Heap allocations via the `box` keyword.
 187     fn box_alloc(
 188         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 189         dest: PlaceTy<'tcx, Self::PointerTag>,
 190     ) -> InterpResult<'tcx>;
 191
 192     /// Called to read the specified `local` from the `frame`.
 193     #[inline]
 194     fn access_local(
 195         _ecx: &InterpCx<'mir, 'tcx, Self>,
 196         frame: &Frame<'mir, 'tcx, Self::PointerTag, Self::FrameExtra>,
 197         local: mir::Local,
 198     ) -> InterpResult<'tcx, Operand<Self::PointerTag>> {
 199         frame.locals[local].access()
 200     }
 201
 202     /// Called before a basic block terminator is executed.
 203     /// You can use this to detect endlessly running programs.
 204     #[inline]
 205     fn before_terminator(_ecx: &mut InterpCx<'mir, 'tcx, Self>) -> InterpResult<'tcx> {
 206         Ok(())
 207     }
 208
 209     /// Called before a global allocation is accessed.
 210     /// `def_id` is `Some` if this is the "lazy" allocation of a static.
 211     #[inline]
 212     fn before_access_global(
 213         _memory_extra: &Self::MemoryExtra,
 214         _alloc_id: AllocId,
 215         _allocation: &Allocation,
 216         _static_def_id: Option<DefId>,
 217         _is_write: bool,
 218     ) -> InterpResult<'tcx> {
 219         Ok(())
 220     }
 221
 222     /// Called for *every* memory access to determine the real ID of the given allocation.
 223     /// This provides a way for the machine to "redirect" certain allocations as it sees fit.
 224     ///
 225     /// This is used by Miri to redirect extern statics to real allocations.
 226     ///
 227     /// This function must be idempotent.
 228     #[inline]
 229     fn canonical_alloc_id(_mem: &Memory<'mir, 'tcx, Self>, id: AllocId) -> AllocId {
 230         id
 231     }
 232
 233     /// Called when converting a `ty::Const` to an operand (in
 234     /// `eval_const_to_op`).
 235     ///
 236     /// Miri uses this callback for creating per thread allocations for thread
 237     /// locals. In Rust, one way of creating a thread local is by marking a
 238     /// static with `#[thread_local]`. On supported platforms this gets
 239     /// translated to a LLVM thread local for which LLVM automatically ensures
 240     /// that each thread gets its own copy. Since LLVM automatically handles
 241     /// thread locals, the Rust compiler just treats thread local statics as
 242     /// regular statics even though accessing a thread local static should be an
 243     /// effectful computation that depends on the current thread. The long term
 244     /// plan is to change MIR to make accesses to thread locals explicit
 245     /// (https://github.com/rust-lang/rust/issues/70685). While the issue 70685
 246     /// is not fixed, our current workaround in Miri is to use this function to
 247     /// make per-thread copies of thread locals. Please note that we cannot make
 248     /// these copies in `canonical_alloc_id` because that is too late: for
 249     /// example, if one created a pointer in thread `t1` to a thread local and
 250     /// sent it to another thread `t2`, resolving the access in
 251     /// `canonical_alloc_id` would result in pointer pointing to `t2`'s thread
 252     /// local and not `t1` as it should.
 253     #[inline]
 254     fn adjust_global_const(
 255         _ecx: &InterpCx<'mir, 'tcx, Self>,
 256         val: mir::interpret::ConstValue<'tcx>,
 257     ) -> InterpResult<'tcx, mir::interpret::ConstValue<'tcx>> {
 258         Ok(val)
 259     }
 260
 261     /// Called to initialize the "extra" state of an allocation and make the pointers
 262     /// it contains (in relocations) tagged.  The way we construct allocations is
 263     /// to always first construct it without extra and then add the extra.
 264     /// This keeps uniform code paths for handling both allocations created by CTFE
 265     /// for globals, and allocations created by Miri during evaluation.
 266     ///
 267     /// `kind` is the kind of the allocation being tagged; it can be `None` when
 268     /// it's a global and `GLOBAL_KIND` is `None`.
 269     ///
 270     /// This should avoid copying if no work has to be done! If this returns an owned
 271     /// allocation (because a copy had to be done to add tags or metadata), machine memory will
 272     /// cache the result. (This relies on `AllocMap::get_or` being able to add the
 273     /// owned allocation to the map even when the map is shared.)
 274     ///
 275     /// Also return the "base" tag to use for this allocation: the one that is used for direct
 276     /// accesses to this allocation. If `kind == STATIC_KIND`, this tag must be consistent
 277     /// with `tag_global_base_pointer`.
 278     fn init_allocation_extra<'b>(
 279         memory_extra: &Self::MemoryExtra,
 280         id: AllocId,
 281         alloc: Cow<'b, Allocation>,
 282         kind: Option<MemoryKind<Self::MemoryKind>>,
 283     ) -> (Cow<'b, Allocation<Self::PointerTag, Self::AllocExtra>>, Self::PointerTag);
 284
 285     /// Called to notify the machine before a deallocation occurs.
 286     fn before_deallocation(
 287         _memory_extra: &mut Self::MemoryExtra,
 288         _id: AllocId,
 289     ) -> InterpResult<'tcx> {
 290         Ok(())
 291     }
 292
 293     /// Return the "base" tag for the given *global* allocation: the one that is used for direct
 294     /// accesses to this static/const/fn allocation. If `id` is not a global allocation,
 295     /// this will return an unusable tag (i.e., accesses will be UB)!
 296     ///
 297     /// Expects `id` to be already canonical, if needed.
 298     fn tag_global_base_pointer(memory_extra: &Self::MemoryExtra, id: AllocId) -> Self::PointerTag;
 299
 300     /// Executes a retagging operation
 301     #[inline]
 302     fn retag(
 303         _ecx: &mut InterpCx<'mir, 'tcx, Self>,
 304         _kind: mir::RetagKind,
 305         _place: PlaceTy<'tcx, Self::PointerTag>,
 306     ) -> InterpResult<'tcx> {
 307         Ok(())
 308     }
 309
 310     /// Called immediately before a new stack frame gets pushed.
 311     fn init_frame_extra(
 312         ecx: &mut InterpCx<'mir, 'tcx, Self>,
 313         frame: Frame<'mir, 'tcx, Self::PointerTag>,
 314     ) -> InterpResult<'tcx, Frame<'mir, 'tcx, Self::PointerTag, Self::FrameExtra>>;
 315
 316     /// Borrow the current thread's stack.
 317     fn stack(
 318         ecx: &'a InterpCx<'mir, 'tcx, Self>,
 319     ) -> &'a [Frame<'mir, 'tcx, Self::PointerTag, Self::FrameExtra>];
 320
 321     /// Mutably borrow the current thread's stack.
 322     fn stack_mut(
 323         ecx: &'a mut InterpCx<'mir, 'tcx, Self>,
 324     ) -> &'a mut Vec<Frame<'mir, 'tcx, Self::PointerTag, Self::FrameExtra>>;
 325
 326     /// Called immediately after a stack frame got pushed and its locals got initialized.
 327     fn after_stack_push(_ecx: &mut InterpCx<'mir, 'tcx, Self>) -> InterpResult<'tcx> {
 328         Ok(())
 329     }
 330
 331     /// Called immediately after a stack frame got popped, but before jumping back to the caller.
 332     fn after_stack_pop(
 333         _ecx: &mut InterpCx<'mir, 'tcx, Self>,
 334         _frame: Frame<'mir, 'tcx, Self::PointerTag, Self::FrameExtra>,
 335         _unwinding: bool,
 336     ) -> InterpResult<'tcx, StackPopJump> {
 337         // By default, we do not support unwinding from panics
 338         Ok(StackPopJump::Normal)
 339     }
 340
 341     fn int_to_ptr(
 342         _mem: &Memory<'mir, 'tcx, Self>,
 343         int: u64,
 344     ) -> InterpResult<'tcx, Pointer<Self::PointerTag>> {
 345         Err((if int == 0 {
 346             // This is UB, seriously.
 347             err_ub!(InvalidIntPointerUsage(0))
 348         } else {
 349             // This is just something we cannot support during const-eval.
 350             err_unsup!(ReadBytesAsPointer)
 351         })
 352         .into())
 353     }
 354
 355     fn ptr_to_int(
 356         _mem: &Memory<'mir, 'tcx, Self>,
 357         _ptr: Pointer<Self::PointerTag>,
 358     ) -> InterpResult<'tcx, u64>;
 359 }